Search

This research focuses on addressing methodological issues that impact the performance of spatially explicit discrete choice agent-based land use models that are estimated with remotely sensed data. The empirical setting considers land use transitions between agroforests, perennial crops, grass and corn, and fallow lands during the period 1984 - 2006 in a Mexican coffee growing region in which relatively high deforestation rates were observed. As a starting point, a Mixed Conditional - Multinomial Logit model is implemented to highlight assumptions and limitations associated with this standard modeling approach. The results indicate that this model produces theoretically inconsistent parameter estimates for the revenue variable associated with three out of four land uses considered in the analysis. To investigate whether those counterintuitive marginal effects are generated from misclassified land use data, a Latent Multinomial Logit (LMNL) model is implemented. This approach allows the identification of land use observations that have a high likelihood of being wrongly classified. A reconfiguration of the dataset based on the LMNL model increased the magnitudes of the marginal effects of the analyzed land use drivers in the theoretically expected directions. Next, because static land use models require limiting assumptions that potentially oversimplify the behavioral process followed by landowners, a structural dynamic discrete choice model of land use decisions is implemented under the assumption that land managers are forward-looking and act to maximize their discounted flow of current and future expected utility within a stochastic environment. A comparison between static and dynamic models shows that the directions of the marginal effects corresponding to time-invariant parcel-specific variables generally have the expected directions independent of the selected modeling approach. More importantly, the marginal effect estimates for the revenue variables of the agroforestry and perennial crops categories have the expected direction in the dynamic model. By contrast the myopic modeling approaches generate counter-intuitive results for the revenue variable that corresponds to perennial crops production, which affects the validity of those results for policy design. Finally, a policy simulation exercise shows the sensitivity of welfare estimates to the discount factor selected as representative of the true value used by decision makers.

The increasing severity and frequency of environmental climate-driven forest disturbances challenge the sustainable management and protection of forests in the United States (U.S.), and the benefits they provide to individuals and communities. Between 1985 and 2015, about 9 million acres of U.S. forest land was disturbed annually across the country. In the western U.S., wildfires, droughts, timber harvesting, and bark beetle outbreaks are the main drivers of tree mortality, loss of recreation opportunities, and water quantity. An understanding of the environmental and socioeconomic consequences of forest disturbances could improve the management and protection of public lands.This study provides monetized estimates of forest disturbances for two key areas of value: recreation and carbon storage. We estimate the monetary value of these ecosystem services and how they are affected by fire, drought or disease due to climate change. We employ benefit function transfer to estimate the economic value of forest ecosystem services u2013 both market and non-market values. This information may improve stakeholdersu2019 understanding of how heterogeneous forests disturbances affect economic and environmental processes at different temporal and spatial scales. The results could also inform the design of spatially targeted forest conservation, pest control, or fire suppression strategies to generate greater benefits to society.

This study presents the first application of deep learning techniques in capturing long-term, time-continuous forest cover dynamics at a continental scale. We developed a spatially-explicit ensemble model for projecting Australia's forest cover change using Long Short-Term Memory (LSTM) deep learning neural networks applied toa multi-dimensional, high-resolution spatiotemporal dataset and run on a high-performance computing cluster.We further quantified the influence of explanatory variables on the spatiotemporal dynamics of continental forest cover. Deep learning greatly outperformed a state-of-the-art spatial-econometric model at continental,state, and grid-cell scales. For example, at the continental scale, compared to the spatial-econometric model, the deep learning model improved projection performance by 44% (root-mean-square error) and 12% (pseudo R squared).The results illustrate the robustness and effectiveness of the LSTM model. This work provides a reliable tool for projecting forest cover and agricultural production under given future scenarios, supporting decision-making in sustainable land development, management, and conservation. Refereed/Peer-reviewed

Recent rates of deforestation on private lands in Australia rival deforestation hotspots around the world, despite conservation policies in place to avert deforestation. This study uses causal impact estimation techniques to determine if a controversial conservation policy-the Vegetation Management Act (VMA)-has successfully reduced deforestation of remnant trees in the Brigalow Belt South, a 21.6 Mha biodiversity hotspot in Queensland. Weuse covariate matching to determine the regulatory effect of the policy on deforestation rates over time, compared to two counterfactual scenarios representing upper and lower estimates of policy impact. The VMA significantly reduced the rate of remnant deforestation in the highest impact scenario, saving 17, 729 +/- 1733 ha during 2000-2016. In the lowest scenario, 'panic clearing' before and after enactment of the VMA minimized the amount of remnant forests saved and may have marginally increased deforestation relative to the counterfactual (-404 +/- 617 ha). At peak effectiveness, the VMA successfully counteracted the amount of remnant deforestation during 2010-2012, but this only represents 4.78% of the 371, 252 ha of remnant forests cleared in the bioregion since enactment in 1999. Thus, while deforestation rates in the region have substantially reduced since the policy was enacted, our results of positive yet limited direct regulatory impact suggests the policy's effectiveness is strongly confounded by other deforestation drivers, like changing socio-economic or climate conditions, as well as new social signals provoked by the policy. The mechanisms through which the policy influences deforestation behavior must be further investigated to ensure real, desirable change is achieved. Refereed/Peer-reviewed

Environmental policies and regulations have been instrumental in influencing deforestation rates around the world. Understanding how these policies change stakeholder behaviours is critical for determining policy impact. In Queensland, Australia, changes in native vegetation management policy seem to have influenced land clearing behaviour of landholders. Periods of peak clearing rates have been associated with periods preceding the introduction of stricter legislation. However, the characteristics of clearing patterns during the last two decades are poorly understood. This study investigates the underlying spatiotemporal patterns in land clearing using a range of biophysical, climatic, and property characteristics of clearing events. Principal component and hierarchical cluster analyses were applied to identify dissimilarities between years along the political timeline. Overall, aggregate landholders' clearing characteristics remain generally consistent over time, though noticeable deviations are observed at smaller regional and temporal scales. While clearing patterns in some regions have shifted to reflect the policy's goals, others have experienced minimal or contradictory changes following regulation. Potential 'panic' or 'pre-emptive' effects are evident in the analysis, such as spikes in clearing for pasture expansions, but differ across regions. Because different regions are driven by different pressures, such as land availability and regulatory opportunity, it is imperative that the varying spatial and temporal behavioural responses of landholders are monitored to understand the influence of policy and its evolution. Future policy amendments would benefit from monitoring these regional responses from landholder s to better assess the effectiveness of policy and the potential perversities of policy uncertainty. Refereed/Peer-reviewed

The shift from net forest loss to gain—forest transition—has been associated variously with economic development, market-driven reforestation, forest policy, and globalization. Evidence shows that governments can expedite forest transition, although economic and institutional failures can distort policy incentives. This study addresses the paucity of spatially explicit empirical research on the robustness of the forest transition hypothesis in a developed country context and identifies factors that may hasten, delay, or even reverse forest transition. We applied spatial-econometric analysis to high-resolution forest cover, climatic, socioeconomic, physiographic, and State-jurisdiction data for the Australian intensive agricultural zone from 1988 to 2014. While environmental and physiographic factors explained the spatial distribution of forests, net forest cover change was significantly associated with trends in farm-output prices inducing deforestation in Queensland, the State with less effective land clearance regulations. Changes in land clearing regulations in Queensland were significantly associated with the national forest cover trends that resulted in forest transition in Australia around 2008. Yet when land clearing regulations and their enforcement were subsequently relaxed in 2012, significant forest cover loss was once again observed in that State, particularly in remnant forests. We conclude that if forest regulatory protection is not effective, net forest loss could resume or increase, even in developed countries, in response to growing incentives for forest conversion to agriculture. Refereed/Peer-reviewed

Forest cover gains and losses occur in response to complex environmental and anthropogenic pressures. Yet the impact of forest gains and losses on the provision of ecosystem services differs markedly. Here we investigate the social costs of potential forest carbon change in Australia’s intensive agricultural region from 2015 to 2050 using spatial forest cover change and forest carbon models combined with climate and socioeconomic projections. More than 24,000 possible scenarios were used to identify the trend and lower and upper bounds of forest cover/carbon change. Net deforestation (3.5 million hectares, Mha) under the lower bound forest cover (LBFC) projection was around one-third less than net reforestation (4.8 Mha) under the upper bound forest cover (UBFC) projection by 2030. However, the CO2 emissions (1.3 Gigatons of CO2, GtCO2) from deforestation were more than double the sequestration (0.5 GtCO2) from reforestation. The social costs (up to 134 billion dollars) of the LBFC were almost five times the benefits of the UBFC (up to 28 billion dollars). The asymmetry decreased over time but persisted to 2050. This shows the markedly different social costs of potential forest carbon losses and gains under global change, evidence which can be useful to policymakers, stakeholders, and practitioners. Refereed/Peer-reviewed